Report No. 21394

World Bank Assistance to CFA CountriesAn Evaluation of Selected Social, Economic and Regional Aspectsof the Bank's Performance

November 9, 2000

Operations Evaluation Department

Document of the World Bank

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Abbreviations and AcronymsAfDB African Development BankAFRISTAT African Institute of StatisticsAIDS Acquired Immune Deficiency SyndromeAPDF African Project Development FacilityBCEAO Banque Centrale des Etats de l'Afrique de l'Ouest [Bank of Western African States]BEAC Banque des Etats de l'Afrique de Centrale [Bank of Central African States]CAEMC Central African Economic and Monetary CommunityCAMC Central African Monetary CommunityCAR Central African RepublicCAS Country Assistance StrategyCDF Comprehensive Development FrameworkC/ESW Country / Economic & Sector WorkCFA "Communaute Financiere Africaine" [African Financial Community]CFAF CFA FrancCIMA Conference Interafricaine des marches d' assurancesCIPRES Conference Interafricaine dePrevoyance Sociale (Inter African Conference for Social Security [14 CFA

countries])COBAC Central African Banking CommisssionEC European CommunityECOWAS Economic Community of Westem African StatesE / HIPC Enhanced I Initiative for Heavily Indebted Poor CountriesESW Economic Sector WorkEU European UnionFF French FrancEV Education VolunteersFIAS Foreign Investment Advisory ServiceFSAL Financial Sector Adjustment LoanGDP Gross Domestic ProductGNP Gross National ProductHIV Human Immune /Deficiency VirusIBRD International Bank for Reconstruction & DevelopmentIDA International Development AssociationIFC International Financial CorporationIMF International Monetary FundMIGA Multi Lateral Investment Guarantee AgencyNGO Non-Governmental OrganizationODA Official Development AssistanceOECD Organization for Economic Cooperation and DevelopmentOED Operations Evaluations DepartmentOHADA Organization pour l'Harminizationdu Droit des Affaires (Regional Organization for Business Law Reform)PPM Pharmacie Populaire du MaliPREM Poverty Reduction and Economic ManagementPRSP Poverty Reduction Strategy PaperQAG Quality Assurance GroupREER Real Effective Exchange RateSSA Sub Saharan AfricaSTABEX Stabilization of Primary Commodities Export Earnings, scheme operated by EUSYSCOA West African Accounting FrameworkUDEAC Central African Customs and Economic UnionUEMOA Union Economique et Monetaire Ouest AfricanineUNAIDS Joint United Nations Program on AIDS / HIVUNICEF United Nations Children's FundWAEMU Western African Economic and Monetary UnionWAMU Western African Monetary UnionWDR World Development ReportWHO World Health Organization

Director General, Operations Evaluations . Mr. Robert PicciottoActing Director, Operations Evaluations Department Mr. Gregory K. IngramManager, OEDCR : Mr. Ruben LamdanyTask Manager : Mr. Louis Goreux

The World BankWashington, D.C. 20433

U.S.A.

Office of the Director-GeneralOperations Evaluation November 9, 2000

MEMORANDUM TO THE EXECUTIVE DIRECTORS AND THE PRESIDENT

WORLD BANK ASSISTANCE TO CFA COUNTRIES

With a population of 100 million, the CFA zone consists of 14 countries sharing a commoncurrency, the CFA franc at fixed parity with the French franc, which induced a monetary and fiscaldiscipline from which CFA countries benefited. From 1970 to 1985, growth in per capita GDP washigher in the CFA zone than in its comparator - a group of 28 countries of Sub-Saharan Africa referredto below as non-CFA countries. A draft of this evaluation was distributed to CODE in July 2000. Thereport is now being re-issued for the purpose of public disclosure.

When the terms of trade deteriorated sharply in the second half of the 1980s, most non-CFAcountries devalued their currencies while the CFA parity with the French franc remained unchanged.The CFA franc became overvalued, which led to an erosion of the competitive position of the zone and tothe 1987-93 depression. The postponement of the devaluation was detrimental to CFA countries whichsuffered a decline in their per capita GDP more severe than non-CFA countries.

The currency was devalued from CFAF50 to CFAF100 per French franc in January 1994 and, inthe five years following the devaluation, annual GDP growth was higher in CFA than non-CFAcountries. But, in spite of the post-devaluation recovery, per capita GDP in CFA countries is hardlyhigher today than it was in 1970. Poverty in Cameroon and C6te d'Ivoire remains much morewidespread today than it was in 1985; in five other CFA countries, half of the population lives below thepoverty line.

The CFA Devaluation: A confidential study, prepared by the Africa Region in March 1988,concluded that a 50 percent devaluation of the CFA franc was needed to restore the competitive positionof the zone. It was argued that, pending the devaluation, the Bank should refrain from extendingadjustment lending on IBRD terms. The Bank reduced its lending to CFA countries before thedevaluation, but did not stop lending because it was caught in a dilemma. On the one hand, as was thecase, not lending would have implied large net transfers from these countries precisely when they weresuffering heavy terms of trade losses. On the other hand, lending US$1.5 billion to C6te d'Ivoire andCameroon on IBRD terms in FY86-90 was not right either. The projects did not turn out well; and theoutcome was rated unsatisfactory for 73 percent of the committed amounts.

The Bank devoted sizable resources to identify the measures needed to support the expecteddevaluation. This analytical work was relevant and it was conducted in difficult conditions. It had to bedone confidentially, but some leakage occurred which, occasionally, created tensions with membercountries. Ultimately, the Bank succeeded in building a solid partnership with the IMF, France and otherkey players in the CFA zone. Once the decision to devalue had been reached by the key donors, it had tobe approved by the Heads of State of all CFA countries, and none of them wanted the devaluation tooccur shortly before presidential elections. In hindsight, it is clear that delaying the devaluation was verycostly in economic and social terms. But the Bank was not responsible for the postponement of thedevaluation and it is not clear that a successful devaluation could have occurred earlier if the Bank hadacted differently.

2

The devaluation occurred in January 1994. It was delayed, but it was well prepared. For thezone as a whole, the annual growth rate of real GDP per capita turned from minus 2.7 percent in 1987-93to plus 1.9 percent in 1994-98. Right after the devaluation, Cameroon, Congo and C6te d'Ivoire weredeclared eligible for IDA credits only, and the undisbursed parts of IBRD adjustment loans werecancelled. IDA was able to disburse US$150 million to CMte d'Ivoire in the month following thedevaluation and the Bank has to be credited for this timely intervention which contributed to the successof the devaluation.

The devaluation was successful but, by now, most of its direct benefits have been cashed in andthe situation of most CFA countries remains fragile. Average per capita GDP of the zone remainedstagnant in 1999. Political and social instability is part of the cause, as is the dependence on a fewprimary commodities. Uneven economic reform and fluctuating external assistance play a role as well.Expressed as percentage of GNP, net external transfers from all sources to the CFA zone doubled from1990-93 to 1994, which was the reward for the devaluation. But they fell in 1995-98 one-third belowtheir 1990-93 level. Fortunately, most CFA countries may become eligible to the enhanced debtreduction facility, which could make a difference. Continued monitoring of its impact is necessary toensure sustainability. To this end, the Bank should assist in the development and collection of bettereconomic and social statistics.

Post Devaluation Developments: Rural Poverty increased massively during the depression.Urban poverty, which was not a serious problem before the devaluation, became one in 1994, when realwages fell sharply in the formal sector and increasing numbers of job seekers joined the ranks of theinformal sector. But little is known of what happened during the six years which followed thedevaluation. In C6te d'Ivoire, the only CFA country where the evolution of poverty is documented withhousehold surveys, the proportion of households falling below the poverty line increased from 11 percentin 1985 to 37 percent in early 1995 before declining to 33 percent in 1998. In other countries, e.g.,Cameroon, poverty increased even more dramatically during the depression. From 1993 to 1998, theterms of trade improved in favor of agriculture and agricultural production increased by about 5 percent ayear, which suggests that average agricultural income increased; but there is no household survey to testthis hypothesis.

Poverty assessments were relevant and they drew public attention to the poverty problem; buttheir efficacy was limited because they were not sufficiently action oriented and they were not wellintegrated into the overall country strategy. Since the majority of the poor live in rural areas, a ruraldevelopment strategy should be the corner stone of the poverty reduction strategy papers (PRSP).

Primary Education. Although countries in the western part of the CFA zone (WAEMU) devoteda large share of their budget to education, enrollment rates in primary schools and adult literacy remainlower than in other countries with comparable per capita income. In part, this anomaly reflected the highcost of teachers. Education budgets went to pay teachers, leaving no resources to pay for all othereducational inputs. Until 1993, teachers wages exceeded ten times per capita GDP; the ratio was morethan three times as large in WAEMU as in the rest of Africa. After the devaluation, the ratio declined,which permitted to raise enrollment rates without increased budgets. The Bank was late in focusing onteacher costs, but it recently organized a series of workshops on the issue with representatives of parentassociations, teachers unions and government officials. In the part of the zone located in central Africa(CAEMC), the evolution has been even more disappointing. In Congo (Brazzaville) and Cameroon,which had virtually reached universal primary education in the early 1980s, quality of teachingdeteriorated and enrollments rates declined in the last decade.

3

Basic Health Services. Post-devaluation, the health services picture remains mixed. On the onehand, the introduction of generic drugs made possible by the Bank and other donors was a success, andthe Bank is now launching a vigorous campaign to contain the spread of HIV. On the other hand, thedelivery of basic health services in government centers did not improve in spite of greater externalassistance, and half of Bank commitments to the health sector are presently at risk. The Bank, as manyother official donors, devoted most of its attention to the government sector, while the share of this sectorin the total delivery of health services was declining. The Bank should concentrate its efforts in areaswhere it has comparative advantages over other donors, e.g. establishing appropriate links between theprivate and the public health sub-sectors and between the health sector and the rest of the economy.

Regional Integration: The Bank followed a regional approach to restructure the WAEMUbanking sector in the late 1 980s. This led to a study of the competitive position of the entire CFA zoneand to the preparation of contingency programs to support the expected devaluation. The regionalapproach has become relevant again. After the devaluation, CFA countries decided to harmonize theireconomic policies and each monetary union was transformed into an economic and monetary union(WAEMU and CAEMC). The WAEMU treaty was ratified by all member countries in August 1994 andthe CAEMC treaty in June 1999. The Bank and the Fund, in close cooperation with the EU and France,assisted the CAEMC secretariat in designing the major tariff and tax reforn implemented in 1994. Theyadvised the WAEMU commission in the establishment of a common external tariff (which is in placesince January 2000) and on other aspects of trade policies. They have also assisted the commission inharmonizing indirect taxes, formulating a competition policy and designing a common investment code(which has not yet been ratified). Moreover, IDA and IFC assisted WAEMU in establishing a regionalstock exchange. The Bank is now involved in some sectoral issues; notably, telecommunications andpower interconnections. Sectoral policy reviews at the regional level could be expanded to cover socialsectors. They would provide an opportunity to know what works and what does not in countries withsimilar structures and to understand why.

To follow a more comprehensive approach without additional-or in some cases with shrinkingresources-the Bank has to become more selective. This requires a better allocation of tasks and closercooperation among development partners. To this end, the Bank should be ready to follow the lead ofother institutions in selected fields.

Contents

1. Changing Economic Fortunes IOverview IGrowth Through the 1970s 5Stagnation in WAEMU, High Growth in CAEMC: 1981-86 6Depression in WAEMU and CAEMC: 1987-93 7Post-Devaluation Recovery 1994-98 9

2. Selected Social Issues 13Poverty 13Primary Education 14Basic Health Services 18

3. World Bank Assistance 23Financial Assistance from the World Bank and Other Sources 23The Bank Lending Strategy 23Non-Lending Assistance 27Performance Ratings 27Projects 28World Bank Performance 28

4. Support to Regional Integration 30

5. Conclusions and Recommendations 32

FIGURES IN TEXT

1.1: Index of Deflated Export Prices of CFA Countries, in US$ and CFAF, 1970-98 41.2: GDP Per Capita of WAEMU, CAEMC and CFA Growth Rates (percent per year) 41.3: Distribution of CFA Countries by Level of Real Annual GDP Growth Rate 1991-98 93.1: World Bank Commitment and Net Transfer to the 13 CFA Countries, FY90-99 263.2: Net Transfers from Multilateral Creditors (as percent of GNP), 1985-88 263.3: Net Transfers from all Sources (as percent of GNP), 1985-98 26

TABLES IN TEXT

1.1: Growth in Real GDP Per Capita 1971-98 44.1 ECOWAS: GNP, GDP and Population, 1997-98 30

BOXES

1.1 The CFA Zone 22.1: Why the Cost of Teachers was so High in Cote d'Ivoire 152.2: Education Volunteers (EV) in Senegal 162.3: Generic Drugs and Central Stores: Cote d'Ivoire, Burkina Faso and Mali 19

Louis Goreux was the Task Manager and the author of this report. Oliver Rajakaruna conducted most of thestatistical work; he was assisted by Dinara Seijaparova and Keith Troung. Annex Tables: Al -A8, Cl-C1 7 andTechnical Notes are available upon request.

1. Changing Economic Fortunes

Overview

1.1 The CFA zone consists of 14 countries' with a total population of 100 million.The 14 countries share the same currency, the CFAF,2 which is at fixed parity with theFF. Since 1948, this parity was modified only once, in January 1994, when the currencywas devalued from CFAF50 to CFAF100 per FF. The creation of the euro five yearslater did not affect the parity between the CFA franc and the French franc; the CFA francbecame linked to the euro which is equivalent to CFAF656 and FF6.56.

1.2 The CFA arrangement differs from usual currency boards in several respects.First, CFA countries have to keep at least 65 percent of their foreign exchange reserveswith the French Treasury. In exchange, the latter provides an overdraft facility which is,in principle, unbounded but, in practice, is not since monetary policies have to betightened when the reserve position deteriorates. Second, with 13 countries sharing thesame currency, a devaluation was a complex operation requiring the consensus of 13Heads of State. Third, when the devaluation occurred, CFA countries decided toharmonize their economic policies and monetary unions became economic and monetaryunions.

1.3 The CFA arrangement induced a monetary and fiscal discipline from which CFAcountries benefited until the mid-1980s; from 1970 to 1985, growth of per capita GDPwas higher in CFA than non-CFA countries.3 When the terms of trade deteriorated in thesecond half of the 1980s, the parity with the French franc remained unchanged and theCFA franc became overvalued. It led to the erosion of the competitive position of thezone and the 1987-93 depression was more severe in CFA than non-CFA countries. CFAcountries were then penalized by the delay in adjusting the currency. During the post-devaluation period (1994-98), GDP growth was higher in CFA than non-CFA countries.But, in spite of the post-devaluation recovery, poverty remains much more widespreadtoday than it was in 1985.

l The analysis covers only 13 countries. It excludes Guinea-Bissau which joined the CFA zone in 1997.2 FCFA for "Franc de la Commnunaute Financiere Africaine" in WAEMU and "Franc de la CooperationFinanciere en Afrique Centrale" in CAEMC. CFAF: African Financial Community Franc, or Central AfricanFinancial Cooperation Franc. Comoros is not covered in this study, its currency is the "franc comorien"(CM) with a fixed parity of CM75 per FF since 1994.3 Nigeria and South Africa were excluded to avoid distorting group averages. Eritrea, Liberia, Namibia,Somalia and Sudan were also excluded for lack of reliable data.

2

Box 1.1: The CFA Zone

The zone is divided in two groups. The eight countries of the Western African Economic andMonetary Union (WAEMU) share a common central bank (BCEAO). The six other countries, whichconstitute the Central African Economic and Monetary Community (CAEMC), share another central bank(BEAC). Each central Bank issues its own currency and the two currencies are not convertible between eachother, although each is convertible with the French franc at the same parity rate.

In January and March 1994, treaties transforming the monetary unions into economic and monetaryunions were signed; acronyms WAMU and CAMC were then replaced by WAEMU and CAEMC. TheWAEMU treaty was ratified in August 1994, but the CAEMC treaty was ratified only in June 1999. Thenumber of CFA countries increased from 11 to 12 when Mali joined WAEMU in 1984, from 12 to 13 whenEquatorial Guinea joined the CAEMC in 1985 and from 13 to 14 when Guinea-Bissau joined the WAEMUin 1997. In assessing the special features of the CFA zone, a comparison is made between 13 CFA countriesand 28 non-CFA countries in Sub-Saharan Africa (SSA). Guinea-Bissau is excluded from the first groupbecause it joined the CFA zone too recently. Seven countries are excluded from the second group: SouthAfrica and Nigeria which account together for 40 percent of SSA's GDP are excluded to avoid distortinggroup averages; Eritrea, Liberia, Namibia, Somalia and Sudan are also excluded due to the lack of reliablestatistics. None of the seven countries excluded received substantial financial assistance from the WorldBank in the 1990s. Since 1994, 36 of the 41 sample countries are eligible only to IDA; 4 countries(Botswana, Gabon, Mauritius and Seychelles) are eligible only to IBRD and Zimbabwe is blend.

In the CFA zone, merchandise exports account for one third of GDP and primary commodities for 90percent of export values. Petroleum exports are insignificant in WAEMU, but they account for about twothirds of total exports in CAEMC and exceed 80 percent in three countries (Congo, Equatorial Guinea andGabon). In 1998, C6te d'Ivoire and Senegal accounted for 59 percent of WAEMU's GDP, while Cameroonand Gabon accounted for 72 percent of CAEMC's GDP. The ratio of per capita GDP between the richestand poorest country in 1998 was 3 to I in WAEMU (Cote d'lvoire versus Niger) against 20 to I in CAEMC(Gabon versus Chad). Moreover, political stability has been greater in WAEMU than in CAEMC where twoIcountries (Congo and CAR) were recently in conflict.

1.4 For the CFA zone as a whole, what was gained in the 1970s was lost during the1987-93 depression and real GDP per capita is hardly higher today than it was thirtyyears ago. This is not much of an achievement, even if the 13 CFA countries did slightlybetter than the 28 non-CFA countries (Table 1.1). This poor performance was partly dueto high population growth. The sharpest fall in per capita GDP from 1980 to 1998 wasrecorded for Niger which became the CFA country with the lowest per capita income.Niger was also the CFA country with the highest annual rate of population growth (3.3percent in 1980-97 and 3.1 percent projected for 1997-2015) and the highest fertility rate(8.1 percent in 1978 and 6.8 percent in 1998).4 At the other end of the spectrum, CMted'Ivoire is the CFA country with the largest decline in fertility (from 7.4 in 1978 to 5.1percent in 1998) and for which the largest reduction in population growth is projected bythe Bank (from 3.2 percent in 1980-97 to 1.7 percent in 1997-2015).

4 A population project in Niger had been approved by the Bank in 1992, but it was closed with 42 percentof the credit cancelled, outcome rated unsatisfactory, sustainability unlikely and institutional developmentimpact negligible. Although the potential for economic growth appears more bleak in Niger than in anyother CFA country, there is no new population project in the pipeline.

3

1.5 The growth pattern reflected to a large extent the fluctuations in export pricesexpressed in CFA franc (Figures 1.1 and 1.2). The main discrepancy between the growthpattern of the two sub-regions occurred in the first half of the 1980s when GDP wasbooming in CAEMC because of favorable oil prices. From 1980 to 1986, real GDP percapita increased by 42 percent in Cameroon while it declined by 13 percent in CMted'Ivoire which was adversely affected by the fall in cocoa prices (Table C7.1). But highgrowth in Cameroon was followed by a dramatic fall. By 1990, GDP per capita inCamneroon was back to its 1980 level; by 1993, it was 14 percent below. Governmentrevenue fell even more steeply, which led to the virtual collapse of govermment services.After having been one of the African countries with the best education and health systemsin the early 1980s, Cameroon became one of the countries with the worst record amongcountries with comparable income levels.

4

Figure 1.1: Index of Deflated Export Prices of CFA Countries in US$ and CFAF,a1970-00 (100=1991-93)

300

2505 0 - - - -- __-_

200

100

50 ___

0

70 72 74 76 78 80 82 84 86 88 90 92 94 96 98 00

^ USS deflated by the MUV G5 Index, CFAF deflated by the CPI for the CFA zone (see Annex Table C6)

Fig. 1.2: GDP per capita of WAEMU, CAEMC and CFA Growth Rates (in percent per year)

WAEMU - ' * CAEMC CFA

10

8 A

6 o AtygoD7-A,8-6 8-3 949 19 90

(1)- (2) -A-o-F . .1 -. . -7.

2 A

-2

-4

-6

-8

70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99

Table 1.1: Growth in Real GDP per capita 1971-98 (average annual percentages)

Country arouo 71-80 87-93 94-98 -71-98 99:Q00(1. 1) WAEMU 0.8 -1.7 -1.4 2.1 -0. 1 1.1(1.2) CAEMC 2.6 2.3 -4.5 1.8 0.4 -0.7(1)=(1.1)-(1.2) CFA 1.7 0.0 -2.8 1.9 0.2 0.4(2) Non-OCFA -0.6 -1.1 -0.9 1.2 -0.5

(1) -(2) CFA-Non-CFA 2.3 1.1 -1.9 0.7 0.7~.... ___

Source: IMFf; See Annex Tables C 7.2 anid C 7.4

5

1.6 Sharp fluctuations in prices of key exports had an adverse effect on long-termgrowth, because in good years governments took advantage of their newly gainedcreditworthiness to borrow more from abroad. When export prices fell, governmentscould not borrow anymore, had to cut spending, and eventually began to accumulatearrears. In order to examine the interaction between exogenous factors (worldcommodity prices) and endogenous factors (policy variables), a distinction is madebetween four periods: (i) high growth in the entire zone through 1980; (ii) stagnation inWAEMU, but rapid growth in CAEMC in 1981-86;5 (iii) depression in the entire zone in1987-93; and (iv) post-devaluation recovery in 1994-98.

Growth through the 1970s

1.7 In the two decades following independence, growth was higher in CFA than innon-CFA countries. Among WAEMU countries, Cote d'Ivoire had the highest growth.It followed an export oriented policy and adopted fairly liberal economic policieswelcoming direct foreign investment and immigrants from neighboring Saheliancountries. By the late 1970s, Abidjan had become a center of attraction for the sub-region. By 1980, per capita GDP was 160 percent higher in C6te d'Ivoire than in Ghana,while Ghana had a higher per capita GDP than CMte d'Ivoire until 1967.6

1.8 The CFA zone benefited from a commodity boom in the mid-1970s, with theindex of export prices (in deflated CFAF) more than doubling between 1972 and 1977(Figure 2.1 and Table C6). Due to the fixed parity with the French franc, the commodityboom did not result in an appreciation of the nominal exchange rate, but it generated aneuphoria which led to investment programs which were far too ambitious with poorlydesigned projects and to support public enterprises which later became money losers.When prices of commodities, other than oil, started to fall in the late 1970s, mostgovernments avoided trimming their investment programs by borrowing frominternational money centers which were anxious to recycle the oil money. Euphoria waseven greater in CAEMC than in WAEMU. Oil production was increasing rapidly inGabon, Congo and Cameroon, while oil prices more than doubled in 1979.

1.9 What did the Bank do during this growth period? Agriculture and infrastructurereceived 83 percent of Bank commitments (Table C 15) and outcome was ratedsatisfactory by OED for 80 percent of the investment projects made in these sectors.Taking all projects approved by the Board by FY80, WAEMU had a better rating thannon-CFA countries for outcome, sustainability and institutional development; CAEMChad the best rating for outcome, but the lowest rating for sustainability and institutionaldevelopment. The Bank impact was limited as it did not succeed in convincing theauthorities to cool down their overheated economies. In 1975, the Bank told CMted'Ivoire that the public investment program was too ambitious; but the remark was notwell received and, when cocoa and coffee prices surged in the next two years, the Bank

5 The period 1980-85 could have been chosen to reflect the turning point in commodity prices. But theperiod 1981-86 was selected to reflect the lagged impact of price changes on GDP growth.World Bank Atlas method.

6

was criticized for being too pessimistic. The IMF Article IV consultation mission to C8ted'Ivoire advised the authorities to sterilize part of the windfall gain from the commodityboom in order to contain inflation, but the government ignored the advice.

Stagnation in WAEMU, High Growth in CAEMC: 1981-86

1.10 The expectation of a strong commodity price recovery provided the rationale forheavy external borrowing in the late 1970s.7 But prices did not recover and the burden ofthe rapidly growing external debt was compounded by the sharp increase in interest rates,which led to the debt crisis.8 In order to service their external debt, governmentsaccumulated domestic arrears. Following Fund and Bank advice, Cote d'Ivoireannounced its intent of rescheduling its debt due to Paris and London Club creditors inDecember 1983. By 1986, all WAEMU countries other than Burkina Faso hadrescheduled their external debts after adopting adjustment programs supported by theIMF.

1.11 The Bank responded to the crisis by increasing its financial assistance and shiftingfrom investment to adjustment lending. Annual Bank disbursements to WAEMUcountries quadrupled from 1975-81 to 1982-84 with two-thirds of the 1982-84disbursements in the form of adjustment lending, compared with less than one-fifth in therest of Sub-Saharan Africa. The rationale behind this shift was that even well conceivedprojects would not flourish in a poor policy environment and that policy changes wereneeded.

1.12 The adjustment programs supported by the Fund and the Bank contributed tostrengthen budgetary discipline. In Cote d'Ivoire, the surplus of the marketing board wastransferred to the budget starting in 1983 and the deficit of the central government fell to4 percent of GDP in 1983 from 13 percent in 1980. The rescheduling of the external debtin 1984-86 provided some breathing space to the Ivorian economy and GDP annualgrowth turned from minus 2.3 percent in 1983-84 to plus 4.2 percent in 1985-86 (TableC7.1). Senegal rescheduled its Paris Club debt five times over the 1981-86 period, andits fiscal and economic situation improved considerably from 1982-83 (Table C8).9

1.13 For WAEMU as a whole, the annual growth rate of real GDP rose from minus 2.4percent in 1983-84 to plus 4.7 percent in 1985-86 (Table C7.1) and it was thought thatadjustment had worked. But the extent of true structural adjustment was limited; inparticular, restructuring of public enterprises was not successful, although it wassupported by large Bank lending. The resumption of economic growth was largely due totemporary improvements in external factors: a modest increase in export prices, a

' The authorities of WAMU countries were not alone in expecting a price recovery. The Bank forecast wasfor a 15 percent increase in the dollar price of cocoa from 1980 to 1982. But in fact, prices fell by one third.8 Interest rates of most commercial loans were based on the six-month LIBOR, which rose from 6.1 percentin 1977 to 12.1 percent in 1979 and peaked at 16.9 percent in 1981.9 From 1982/83 to 1986/87, the budget deficit (before grants) fell from 8.8 percent to 2.6 percent; theexternal current account deficit (before grants) fell from 18.4 percent to 11.3 percent and inflation (asmeasured by consumer prices) fell from 14.5 percent to 0.2 percent. These improvements were associatedwith higher growth. Annual average GDP growth rose from minus 0.7 percent in 1983/84 and 1984/85 toplus 4.2 percent in 1985/86 and 1986/87 (Table C8).

7

depreciation of the French franc in relation to the US dollar through early 1985 and anincrease in external assistance, particularly through rescheduling of the external debt.

1.14 CAEMC. High Growth through 1985: The first half of the 1980s was a great timefor oil countries. With the rapid expansion of oil production and favorable prices, thecombined GDP of Cameroon and Congo (Brazzaville) increased at 10 percent a year onaverage. From 1979 to 1985, government oil revenues expressed in current CFA francsincreased ten fold in Congo. With such a strong increase in oil revenues, the oil countriesdid not need budgetary assistance and were not receptive to Bank policy advice.

Depression in WAEMU and CAEMC: 1987-93

1.15 During the first half of the 1980s, the oil countries in CAEMC had everythingworking for them. But, in 1986-89, the same countries had everything working againstthem (stagnating oil production, falling dollar prices for their exports and sharpappreciation of the CFA franc in relation to the dollar). With oil prices falling by 55percent in dollar terms from the last quarter of 1985 to the second quarter of 1986, Congoand Gabon adopted a Fund supported program and rescheduled their Paris Club debtbefore the end of 1986, while Cameroon waited for two more years before doing thesame. The annual GDP growth of CAEMC fell from plus 6.8 percent in 1984-85 tominus 3.3 percent in 1987-88 (Table C7.1).

1.16 In WAEMU, the 1985-86 gains were short-lived. A strong appreciation of theCFA franc was aggravated by the fall in export prices valued in dollars. In May 1987,COte d'Ivoire suspended payments to Paris and London Club creditors and normalrelations with these creditors were not restored until the devaluation. By 1987, the entireCFA zone was in crisis. Within three years (from 1985 to 1988), export earnings in realtermns fell by half in C6te d'Ivoire and two-thirds in Cameroon. The financial situationwas aggravated in Cameroon by a political disobedience campaign linked to the 1992presidential election; government revenues expressed in real terms and on a per capitabasis fell by 60 percent from 1985/86 to 1993/94.

1.17 CFA countries were not alone to suffer from deteriorating terms of trade. WhenIndonesia and Malaysia, as well as neighboring Ghana and Nigeria, felt the squeeze, theyreacted aggressively by depreciating their currencies. But the parity between the CFAand the French francs remained unchanged and the real effective exchange rate of thezone (REER) appreciated while the terms of trade were deteriorating, worsening thecompetitive position of the zone. Tariff and non-tariff barriers could not prevent illegalimports, which became more attractive. Unable to compete with imports, the formalsector was shrinking. In an attempt to offset the reduction in the tax basis, governmentsraised tax rates, which drove more enterprises to the informal sector.

1.18 Public revenues were shrinking, but the wage bill was not. As a result, the publicwage bill rose in Cameroon from 25 percent of government revenues in 1985/86 to 63percent in 1992/93; for the WAEMU as a whole, the public wage bill reached 69 percentof government revenues in 1993. As treasuries became unable to borrow, non-wageexpenditures were cut to the bone and governments accumulated payments arrears.Government's suppliers which were not paid on time became unable to service their

8

loans, which led to a sharp increase in the share of non-performing loans in Bankportfolios. What started as government arrears ended up as losses for the banks,government-owned banks in particular. By the end of 1987, state owned banks hadbecome bankrupt and WAEMU was on the eve of a banking crisis. The most seriousproblems were in CBte d'Ivoire, Senegal and Benin which called the World Bank forhelp.

1.19 Instead of proceeding on a piece meal basis country by country, the World Bankdecided in January 1988 to conduct a study covering the entire sub-region. After findingthat non-performing loans in the sub-region accounted already for one fourth of totalcredits, the Bank proceeded to a thorough analysis of the factors responsible for theaccumulation of non-performing loans. The study concluded that new money should notbe injected in the system before implementing a number of specified changes in creditpolicies and bank regulations. The Bank study was used as the basis for thorough policydiscussions at the sub-regional level with IMF, BCEAO and France, before coming toimplementation at the national level. Most of the reforms initially recommended by theBank were adopted by the seven Heads of State in September 1989. The FinancialStructural Adjustment Loan (FSAL) with Senegal was cofinanced by France and theUnited States; it became effective in December 1989, while the FSAL with COte d'Ivoirebecame effective only twenty-seven months later.'10 In spite of this delay, the health ofthe WAEMU banking sector had been restored before the devaluation. This was not thecase in CAEMC; an attempt to restructure the banking system of Cameroon was made inthe early 1990s, but it was unsuccessful.

1.20 The WAEMU banking study was the first important study conducted by the Bankat the regional level and it led to an assessment of the competitive position of the entireCFA zone. But, in view of the sensitivity of the parity issue, the new study had to beconducted with great discretion. A confidential study conducted by the Africa Regionconcluded in March 1988 that a devaluation was unavoidable and that the parity had to bechanged from CFAF50 to CFAF100 per FF. In an internal memorandum, it was arguedthat the Bank should refrain from extending budgetary assistance before the devaluation,since the cost of servicing a Bank loan would double in CFA francs after the devaluationwhile government revenues would not increase as much. The issue was of particularimportance to the four countries not eligible to IDA, which accounted together for overhalf of the combined GDP of the CFA zone. This recommendation could not beimplemented and in 1989, IBRD committed US$150 million to Cameroon and US$250million to Cote d'Ivoire in the form of adjustment lending.

1.21 The authorities of many CFA countries were initially fearful of a devaluationwhich was visualized as a jump in the unknown. But perceptions evolved progressivelyas it became clear that the "internal adjustment strategy" was not working. In particular,an increasing number of governments were becoming unable to pay their civil servantson time, which was a source of civil unrest. Finding a date for the devaluation whichwould be acceptable to each of the 13 countries was not easy, since Heads of State did

"' The outcomes of the two FSALs were rated as satisfactory by OED, although the common wisdom was toprovide a financial adjustment loan only when the macroeconomic framework was satisfactory, which wasnot the case since the exchange rate was overvalued.

9

not want the devaluation to occur shortly before presidential elections Presidentialelections were held during the last quarter of 1992 in Cameroon and during 1993 inSenegal and Gabon. Then, Ivorian President Houphouet Boigny fell sick and died inDecember 1993.

Post-Devaluation Recovery 1994-98

1.22 The devaluation occurred in January 1994. It was delayed, but it was wellprepared. Since the immediate effect of the devaluation was to double the cost ofservicing the external debt in CFA francs, additional debt relief was needed quickly;without it, the servicing of the public external debt would have exceeded 80 percent ofgovermment revenues in half of the countries. Within the two weeks which followed thedevaluation, France cancelled the remaining part of its ODA debt to the poorest CFAcountries and half of it to the others. Within the three months following the devaluation,six CFA countries regeheduled their debt to the Paris Club under special terms. Por theyear 1994 as a whole, the 13 CFA countries obtained a debt reduction equivalent to 22percent of their combined GDP. In addition, they received special financing assistanceequivalent to 4.9 percent of GDP (of which 2.2 percent from the Bank, 1.3 percent fromFrance and 1.2 percent from the Fund), and project aid equivalent to 4.2 percent of GDP.

Figure 1.3: Distribution of CFA Countries by Level of Real Annual GDP Growth Rate

below 3 percent, between 3 Rnd 5 pereent, and Ab6ve s pereent, 1991-99

LXiGwth < 3% 0Growth> 3% <5% rw 13

Devaluation

I' 1,1

z@ a0~~~~~~1

1991 1992 1993 1994 1995 19 1997 199

1.23 The CFA devaluation was a success. The macroeconomic objectives which hadbeen set in January 1994 for 1996 were reached for the 2tne Q a whole, with WAEMUdoing a bit better than CAEMC and Cote d'Ivoire doing the best. For the zone as awhole, the annual growth rate of real GDP turned from minus 0.2 percent in 1990-93 to

10

plus 4.4 percent in 1994-98 (Figure 1.3)." The improvement was not limited to GDPgrowth. It affected all key macroeconomic indicators and provided a text book exampleof what can be achieved with a successful devaluation (Table C9).

1.24 The CFA devaluation was successful bccause it had been thoroughly prepared.Contingencies plans had been drawn up far in advance and updated year after year. Mostof the measures which had been programmed were applied, even if delays occurred in anumber of countries. In particular, govemments used considerable restraint in raising thenominal wages of public servants which was essential to restore the budget equilibriumand the competitive position of the zone. Because wage increases in the formal sectorwere successfully contained, inflation remained within the original program targets.While the value o0 the PF had doubled in relation to the CFAF in January 1994, the costof living differential with France increased by only 47 percent in WAEMU and 55percent in CAEMC from 1993 to 1997. As a result, the real effective exchange rate(REER) depreciated by about one-quarter, which was the original objective of thedevaluation.

1.25 Profitability improved in several import substitution industries with low importcontent, but the main gains were mn the export sector. With improvements in the dollarprices of commodities other than oil in 1994 and 1995, export prices valued in CFAFmore than doubled, while labor costs increased only moderately. The export sectorbecame very profitable and its share in GDP rose from 25 percent in 1993 to 35 percentin 1995. The supply response was strong and the ratio of domestic savings over CDPalmost doubled in CFA countries from 1991-93 to 1994-98 (Table C9). A number ofpublic enterprises became profitable and were privatized, which had the effect of raisinggovernment revenues, direct private investment and capital inflows. MoQroycr, privatecapital inflows were reversed in several WAEMU countries, because nationals livingabroad took the opportunity to build houses in the CFA zone which generated a strongreal estate recovery in CMte d'lvoire and Senegal. But most capital inflows wereassociated with privatizations.

1.26 The Bank played an important role in promoting privatization by inducinggovernments to create an environment more friendly to the private sector and reducingproduction costs through greater competition. Some steps in this direction had alreadybeen made before the devaluation, especially in Cote d'Ivoire where committees hadbeen set up, with representatives of the public and private sectors, to consider ways ofimproving produetivity. Thin may be one of the reasong why privatization progressedmore rapidly in C6te d'Ivoire than in Cameroon. Throughout the zone, there is a broadrecognition that the Bank played a useful role by helping to create an environment inwhich privatization could proceed in a fair and transparent manner.

1.27 The banking system in Cameroon was successfully restructured in 1996 after thefailed attempt of 1991. By 1997, the banking system of the CFA zone had generally

" Out of 13 CFA countries, the number of those with an annual growth rate in excess of 3 percent rose from2 in 1993 to 10 in 1995 and 12 in both 1996 and 1997. The number of countries with a growth in excess of 5percent rose from I in 1993 to 9 in 1998. The annual growth of per capita GDP was 2.2 percentage pointslower in CFA than non-CFA countries in the three years preceding the devaluation. But, in the five yearsfollowing it, growth was 0.7 percentage point higher in CFA than non-CFA countries.

11

become liquid and profitable. What had been a vicious cycle before the devaluationbecame a virtuous cycle afterward.

1.28 WAEMU initially Benefitedfrom Favorable Exogenous Factors. The rapideconomic recovery in Cote d'Ivoire and, to a lesser extent, in other WAEMU countrieswas partly due to an improvement in terms of trade. From 1993 to 1995, the terms oftrade of CMte d'Ivoire improved by 33 percent and those of WAEMU as a whole by 15percent, while the terms of trade of CAEMC improved by only 2 percent. Moreover,C6te d'Ivoire did not suffer from the increase in energy prices as did Burkina Faso, Maliand Senegal which were highly dependent on oil imports. Half of the Ivorian powerconsumption was generated from hydro, and following recent gas discoveries, theremainder was derived from domestically produced gas. In recent years, Cote d'Ivoirehas been exporting power to neighboring countries. The Togolese economy, which hadsuffered from severe political unrest in 1992 and 1993, strongly recovered in 1994; butthe recovery was interrupted by renewed political problems following the contested 1998presidential elections. Among CFA countries, Cote d'Ivoire had the fastest recovery, butit benefited from favorable exogenous factors and was probably the country best preparedfor the devaluation.

1.29 Recovery was slower in CAEMC than in WAEMU, partly because oil prices fellin the first half of 1994; but CAEMC was catching up with the late 1996 price increase,until oil prices fell sharply in 1998. GDP growth in oil countries was also affected bychanges in the volume of oil production; the extreme case was Equatorial Guinea, acountry with population under half a million, where real GDP more than doubled from1996 to 1998 because of major oil discoveries. The slower recovery in CAEMC alsoreflected the poor quality of governance in several countries and internal conflicts inCAR and Congo.

1.30 The Public Wage Bill Requires Attention. The WAEMU public wage bill fellfrom 69 percent of tax revenue in 1993 to 39 percent in 1997. Public sector employmentincreased only slightly and the average wage of civil servants declined in relation to percapita GDP, while tax revenues grew faster than GDP. The purchasing power of civilservants fell substantially which induced corruption, as civil servants looked for means ofcompensating for the loss in their real wages. The problem was particularly acute in thejudiciary system. It was at its worst in Cameroon, where nominal wages, which had beencut by half in 1993, were not raised before February 1997 in spite of the inflation inducedby the devaluation.

1.31 One lesson which can be drawn from the experience of the CFA countries is thatreducing nominal wages of civil servants can be counterproductive. It was attempted inCote d'lvoire and Senegal, but the measure had to be rescinded soon after being enacted.In Cameroon, cutting nominal wages in 1993 had disastrous effects on the morale and theproductivity of the civil service. Reducing the size of public service would have beenanother option for reducing the public wage bill; but it was politically and sociallydifficult to dismiss public servants in the mist of a depression when urban unemploymentwas increasing. It was possible to reduce real wages substantially in the context of the1994 devaluation, because people were used to a stable currency; but it was a traumaticexperience and a new devaluation may not have the same effect. Several CFA countries

12

were able to reduce the teacher wage bill after the devaluation without generating socialunrest by recruiting new teachers at lower pay and grand-fathering old teachers.

1.32 The Situation is Fragile. Commodity prices are volatile and CFA countriesremain highly dependent on few primary commodity exports. Petroleum and timberaccount for 93 percent of the combined exports of Congo, Gabon and Equatorial Guinea.Cotton fiber, livestock on the hoof and gold account for three quarters of the combinedexport earnings of Burkina Faso and Mali. Cocoa and coffee account for 40 percent ofthe export earnings of Cote d'Ivoire which was adversely affected by the recent fall in theprices of these two commodities. As dollar prices fell by 40 percent for cocoa and 45percent for coffee from December 1998 to March 2000, annual growth in per capita GDPin Cote d'Ivoire fell from 2.1 percent in 1997 to 0.2 percent in 1999 and may even fallfurther in 2000. In CAEMC, the oil price fall in 1998 was compounded by specialproblems in Gabon;'2 as a result, the net foreign assets of the sub-region declined sharplyfrom December 1997 to March 1999 and growth in per capita GDP growth fell from plus2.8 percent in 1997 to minus 2.2 percent in 1999. But, with the sharp oil price increase in1999, CAEMC's per capita GDP growth is expected to improve in 2000.

12 Excessive public spending before the 1998 presidential elections was followed by a sharp decline in 1999,which generated a depression. These developments were aggravated by a progressive decline in oilproduction. which raises serious problems for the medium-term. Real per capita GDP fell by about 8 percentin Gabon in 1999, which partly explains the 2.2 percent decline estimated for the CAEMC average in that year.

13

2. Selected Social Issues

Poverty

2.1 In C8te d'lvoire, where the evolution of poverty is best documented, theproportion of households below the poverty line increased dramatically during thedepression, rising from 11 percent in 1985 to 31 percent in 1993. It rose further to 37percent in early 1995, but declined to 33 percent in 1998. Urban poverty, which was verylimited before 1994, became a problem immediately after the devaluation with onequarter of urban households falling below the poverty line in 1995; but poverty declinedsubstantially in Abidjan from 1995 to 1998.

2.2 During the depression, poverty increased even more dramatically in Cameroonthan in Cote d'Ivoire, because the fall in per capita consumption was steeper. Half of thetotal population fell below the poverty line in 1994 and urban poverty became a majorproblem when nominal wages of civil servants were cut by half in 1993 without beingadjusted for inflation before 1997. As quoted in the 1996 CAS: "While fewer than 1percent of households in Yaounde (the political capital) and Douala (the economiccapital) fell below the poverty line in 1983, more than 20 percent of households inYaounde, and 30 percent in Douala, did so in 1993."'3 In the post devaluation period, theterms of trade improved in favor of agriculture and agricultural production increased byover 5 percent a year. This suggests that poverty declined in rural areas where 84 percentof the poor were located when the devaluation occurred; but, there is no householdsurvey to test this hypothesis.

2.3 In CFA countries other than CMte d'Ivoire and Cameroon, the evolution ofpoverty cannot be documented by comparing household survey findings over time. But,in five of the ten countries for which recent data are available, more than half of thepopulation was below the poverty line. 14 By shifting the terms of trade in favor ofagriculture, the devaluation reduced income disparities between urban and rural areaswhere most of the poor lived before the devaluation. The impact was not, however, thesame in rural areas producing export crops and in those producing traditional foodstufffor domestic consumption, nor in the capital city and secondary towns. But thesechanges are not well documented and, even six years after the event, there is not enoughstatistical information to assess the impact of the devaluation on poverty.

2.4 The Bank's Poverty Assessments. Most poverty assessments in the CFA zonewere of limited operational value and their findings were not generally well integrated inthe overall strategy. Since the majority of the poor live in rural areas, a ruraldevelopment strategy should be the corner stone of the new PRSP. But a number of CFAcountries do not have a rural strategy, and considerable uncertainties remain on the basic

13 Paragraph 6 in "Country Assistance Strategy" 1/17/1996, Report 15275-CM.14 Benin 15 percent (1994), Gabon 23 percent (1997), Togo 44 percent (1996), Burkina Faso 44 percent(1996), Guinea Bissau 49 percent (1994), Senegal 53 percent (1995), Mali 55 percent (1993), CAR 61percent (1994), Niger 67 percent (1996) and Chad 78 percent (1997). The data are not always comparableamong countries; thus, poverty incidence was probably underestimated in Benin because the survey wascarried out during an exceptionally good harvest year.

14

data needed to formulate such a strategy: notably, the percentage of the population livingin rural areas, the importance of migrations and the shares of rural income derived fromagricultural and non-agricultural activities. Among the 40 economic and sector workreports (ESW) issued in FY98 which were reviewed by the Quality Assurance Group(QAG), poverty assessments received the lowest score, with only 43 percent foundsatisfactory against an overall 72 percent average. Moreover, of the four criteria rated byQAG, likely impact had also the lowest score (53 percent).'5

2.5 Conducting consumption surveys among several thousands households is a costlyand lengthy operation which could have been more efficiently exploited. Such surveysprovide the data needed to calculate income-elasticities'6 for various consumption itemsand the techniques for doing it are now well established. The econometric analysis canbe conducted at low additional cost, provided it is programmed in the survey design.Better estimates of elasticity coefficients for food and non-food items would lead tobetter demand projections which are needed to formulate realistic long-term strategies forthe agricultural sector. These coefficients could also be used to estimate changes inpoverty in between large and expensive household surveys. The evolution in livingstandards of specific population groups could thus be derived from changes in theconsumption of items which are highly income-elastic and for which consumption dataare readily available.

2.6 The most effective way of helping the poor is to provide them with betteremployment opportunities, which requires better basic education and health services.CFA countries have to improve their scores in primary education and delivery of basichealth services and the Bank has also to improve the efficacy of its assistance in thesetwo fields.

Primary Education

2.7 WAEMU. C6te d'Ivoire devoted about 40 percent of its public budget and closeto 8 percent of its GDP to education in the 1 980s, which are among the highest ratiosfound anywhere. But, in the early 1990s, only one child out of two was attendingprimary school and one adult out of three was literate. This paradox reflected theabnormally high costs of teachers. In C6te d'Ivoire, the cost of a teacher in publicprimary school was equivalent to 13 times per capita GDP in 1992; if the same ratio wereapplied to the US today, a teacher would cost US$350,000 a year.

2.8 With enrollment in private schools accounting for only 10 percent of totalenrollments in WAEMU, compared with 20 percent in non-CFA countries, the role of theprivate sector remains minor (Tables C3. 1 -C3.5). Public primary education isadministered by the central government and, up to recently, all public school teacherswere civil servants. Since primary education is a labor intensive activity, its cost is

I5 "Quality of ESW in FY98, A QAG Assessment", 12/11/98.16 Elasticities would relate to total expenditures and not to income which cannot be accurately recorded.Thus, consumption of a given item, say rice or sugar (expressed in terms of expenditure or physicalquantity) would be related to total household expenditures. Beer and non-alcoholic beverages are highlyincome elastic and, in Cote d'lvoire, their consumption levels are accurately known on a monthly basiswith a breakdown between the capital city (Abidjan) and the rest of the country.

15

closely linked to the civil service pay scale which is much higher in WAEMU than in therest of Sub-Saharan Africa. In terms of per capita GDP, the average cost of civil servantswas three times that of non-CFA countries in 1993. Such a disparity occurred because,during the thirty years following independence, real wages of civil servants were erodedby inflation much more severely in non-CFA countries than in WAEMU where inflationwas about the same as in France.

2.9 The differential in civil service salary scales was compounded in Cote d'Ivoire bythe "deccrochage" of teachers' salaries (Box 2.1). When the depression came, non-wageexpenditures were cut to the bones, the teachers wage bill exceeded 95 percent of thecurrent cost of primary education and the quality of education suffered. Housingallowances were progressively eliminated and teachers recruited after 1992 wereremunerated according to the normal civil service pay scale ("raccrochage"). In spite ofthe decline in real wages following the devaluation, the wage bill remained too high toraise enrollments substantially and, when the government decided in 1996 to recruitassistant teachers at reduced pay, it received twenty applications for each positionavailable. The gross enrollment rate rose from 65 percent in 1992 to 71 percent in 1996,while the share of the government budget devoted to education fell from 7.8 percent ofGNP in 1987-93 to 5.0 percent in 1994-97.

Box 2.1: Why the Cost of Teachers was so High in Cste d'Ivoire

When Cote d'lvoire became independent, President Houphouet Boigny treated education as "thepriority of priorities" and the education system was designed to build an Ivorian elite comparable to thatof developed countries as rapidly as possible, without regard to costs. In order to shorten thedevelopment path, the country imported a massive number of teachers during the twenty-five yearsfollowing independence. Until 1975, over 90 percent of teachers were non-lvorian. During the 1975-78boom, students were not interested in teaching; they wanted a public enterprise job. When this outletdisappeared with the recession, the government's response was to replace expatriate teachers bynationals. But Ivorian graduates asked for salaries comparable to those earned by the expatriate teachersthey were replacing and most of their demands were met. Teacher salaries became so high that their payscale had to be disconnected from the salary scale of other civil servants ("deccrochage"). In 1979, ahigh school teacher in the second cycle was paid US$ 17,000 at the prevailing exchange rate, which is theequivalent of US$32,000 today. Teachers in CMte d'lvoire were paid virtually as much as inindustrialized countries with per capita income more than ten times higher.

2.10 The case of CMte d'Ivoire was not unique. The cost of a primary school teacherexpressed in terms of per capita GDP was even higher in Burkina Faso than in Coted'Ivoire. In Senegal, universal primary education has been the objective sinceindependence in 1960. The target date to achieve it was initially set at 1975, but thetarget date was repeatedly postponed and it is presently set for 2015. After rising from 43percent in 1970 to 58 percent in 1986, the gross enrollment rate fell to 54 percent in 1993while the share of government expenditures devoted to education rose to a peak of 33percent. Since then, the average cost per teacher was reduced by introducing the assistantteacher position in 1993 and education volunteers in 1995 (Box 2.2); the enrollment raterose from 54 percent in 1993 to 60 percent in 1997/98, while the budget share devoted toeducation did not increase.

2.11 Niger recruited volunteers with IDA support. Burkina Faso created satelliteschools covering the first three years of primary education with locally recruited teachers

16

receiving much lower pay. The experiment conducted with participation of localcommunities was promising and the proportion of girls attending these schools (48percent) exceeded the national average. When governments of WAEMU countriesadvertised contractual teacher positions with remuneration much lower than thosereceived by civil servants, the number of applications far exceeded that of the positionsavailable, which shows that costs of teachers with civil service status was disconnectedfrom the realities of the labor market in the early 1990s.

Box 2.2 Education Volunteers (EV) in Senegal

When the authorities became convinced that enrollment rates could not rise without reducing theaverage cost of teachers, they created in 1993 the assistant teacher position which required lower academicqualifications but carried lower pay. A Bank supported project (PDRH2) specified that assistant teachersshould account for no less than 80 percent of new recruits. But, the most dramatic change occurred in 1995,when the government announced that, during each of the four following years, 1,200 education volunteers(EV) would be recruited under four-year contracts. EV received a CFAF50,000 monthly stipend, whichrepresented only one third of the cost of regular teachers, but local communities were to provide them withhousing facilities. Because most graduates were unable to find jobs at the time, 28 applications werereceived for each available position. Among those recruited, 22 percent had their "brevet" (which isrequired for teachers with civil service status), 65 percent had their "baccalaureat" and 12 percent haduniversity degrees. According to surveys conducted among parents, EV's performances were as good orbetter than those of regular teachers. 1< By 1998/99, EVs accounted for nearly one fourth of the teacherpopulation remunerated by the government and teachers in private schools for another 15 percent.

Public teachers unions were initially strongly opposed to the EV scheme, but the governmentsucceeded in avoiding open conflicts. EVs appreciated a steady employment and the attrition rate was verylow. Most felt, nevertheless, that they were underpaid and many were worried about their future after theexpiration of their four-year contract. This fear was appeased when the government announced that thosewishing to pursue a teaching career could do so.

2.12 The Bank Response. The Bank has been involved with education for a long time.The first education project in the CFA zone was financed in 1968 and educationaccounted for almost 9 percent of total Bank commitments to the zone through 1980.Performance, however, was not particularly good. OED ratings were lower for educationthan for other projects as regards outcome, sustainability and, especially, institutionaldevelopment impact (Table C4). Already in the 1 980s, the Bank advocated a shift ofgovernment resources from tertiary to primary education and an increase in the numberof girls attending schools. In the early 1990s, the Bank emphasized the need to improveefficiency by reducing repeating and drop-out rates and by using more efficiently bothteachers (two grades being taught simultaneously by one single teacher in areas with lowpopulation density) and classrooms (using the same classroom for one set of students inthe morning and another one in the afternoon in areas with high population density). Butthe Bank was late in focusing on the problem of teachers' salaries.

2.13 The seminal study was the 1992 public expenditure review in Burkina Faso,which presented alternative scenarios of the Burkinabe economy through 2011highlighting the need to reduce teacher costs. The critical importance of teacher costs

17 Evaluation by parents in "Etude sur les strategies de mise en place d'un statut perenne pour lesvolontaires de l'education nationale" Ministry of National Education, Republic of Senegal, November1997.

17

expressed in terms of per capita GDP was emphasized by OED in 199718 and later by A.Mingat in a workshop organized by the Bank in February 1998 in Dakar with governmentofficials and task managers of seven WAEMU countries. Similar workshops were heldwith teachers unions in Ouagadougou and with representatives of students' parentassociations in Paris. This participatory approach extended to parents, teachers unionsand government officials provided a better understanding of the nature of the problem bythe various partners.

2.14 CAEMC. The problems in CAEMC were not the same as in WAEMU. Gabon,Congo and Cameroon had already virtually achieved universal primary education in theearly 1980s, when private schools accounted for 35 percent of primary enrollments inCameroon and Gabon. GDP rose rapidly through 1985, supported by significant oildiscoveries, Governments shared part of the oil bonanza by recruiting more civil servants.Before the devaluation, the number of civil servants per thousand inhabitants was twiceas high in CAEMC as in WAEMU; but the public wage bill measured as percentage ofgovernment receipts was hardly higher, because the average cost of civil servants wasseven times GDP per capita in CEAMC instead of 14 in WAEMU.'9

2.15 When the price of oil collapsed in 1986, public revenues fell sharply andgovernments became unable to pay their civil servants on time and to allocate subsidiesto private schools. In Cameroon, where real public wages fell by two-thirds from 1985 to1995, many teachers reduced the number of hours devoted to official duties and took jobson the side to complement their government salaries. The quality of public educationdeteriorated and the gross enrollment rate fell from 112 percent to 81 percent in ten years.An increasing number of teachers in public schools had to be recruited and paid byparents; these teachers now account for some 30 percent of all public school teachersnation-wide and 60 percent in the poorest province. The government attempted topalliate the shortage of teachers by recruiting 1,600 assistant teachers under contract in1996/97 and 3,200 more in 1997/98. The cost of contractual teachers was less than halfof that of teachers with civil service status, but more than twice that of teachers recruitedby parents. In spite of the huge wage differential between the three categories ofteachers, the performance of those with higher pay was no better than that of those withlower pay according to surveys rating parent satisfaction. Bank staff spent three years inpreparing an education project in Cameroon, but it encountered so many problems that itdecided to drop the project in 1998. But Cameroon is now anxious to improve its imageto qualify for the enhanced HIPC facility and the Bank is now supporting a schemeaiming at reducing corruption in the public education sector, which is highly relevanteven if the chance of success is uncertain.

2.16 In CAR, school teachers outside Bangui were rarely paid on time in the threeyears preceding the devaluation and, in the middle of political unrest, primary schools

Ih See note I Annex B.19 In 1993, the average cost of a civil servant was three times per capita GDP in Gabon, 5 times in Congoand six times in Camneroon, while it exceeded 10 in the two non-oil countries (CAR and Chad), raising toseven the CAEMC average. However, since the number of civil servants per thousand inhabitants wastwice as high in CAEMC as in WAEMU, the public wage bills measured as percentage of governmentreceipts were about the same in the two sub-regions. Table 17, page 60, "La Devaluation du Franc CFA,Un Premier Bilan Apres la Devaluation" by L. Goreux, World Bank, December 28, 1995.

18

remained closed for two full academic years (1991/92 and 1992/93). In Chad, which hadthe lowest enrollment rate of the sub-region at the end of an extended war period, thesituation has been improving lately and the education projects supported by the Bank arenow doing well. By contrast, the situation has been sharply deteriorating with the civilwars in Congo, which was the CFA country with the highest enrollment rate in the 1980s.

Basic Health Services

2.17 In the adjustment of the 1980s, the emphasis was placed on reducing the budgetdeficit, which often resulted in lower allocations to the health sector. This led to strongprotests by NGOs and the United Nations Children's Fund (UNICEF) which advocated in1987 "Adjustment with a Human Face" and health becarne a top priority of the donorcommunity in the 1990s. In spite of this new priority and higher economic growth, lifeexpectancy at birth stagnated during the 1990s in Sub-Saharan Africa, while it hadincreased by 2.5 years during the 1980s (Table C5.1). The improvement in the 1980swas largely due to a decline in infant mortality, while the stagnation in the 1990sreflected the increasing prevalence of HIV which reduced the projected life expectancy atbirth. According to World Health Organization (WHO) estimates, with only 10 percentof the world population, Sub-Saharan Africa accounts for 80 percent of AIDS relateddeaths worldwide.

2.18 Although HIV prevalence in CFA countries remains much lower than in SouthernAfrica it is rapidly increasing, notably in CAR, CMte d'Ivoire, Togo, Congo and BurkinaFaso (Table C5. 1). Experience shows that, once rates exceed seven percent as theyalready do in several CFA countries, HIV can spread very rapidly and reach catastrophicproportions within ten years. The epidemic was contained in some countries (such asSenegal and Uganda) because the authorities took the lead in launching a comprehensivecommunication campaign with large scale participation. But this has not yet occurred inseveral CFA countries with high HIV prevalence. The Bank joined the UNAIDS effortsin 1996 and launched a vigorous campaign to contain the HIV spread toward the end of1999. But, in view of the complexity and magnitude of the task, the Bank should workvery closely with the other agencies which can contribute to containing the HIV spread.An example of successful partnership was the introduction of generic drugs in the CFAzone.

2.19 At the time of the devaluation, it was feared that medicinal drugs would becomeunaffordable to the vast majority of the population, since virtually all drugs wereimported.20 To prevent this the European Community, the Bank and other major donorscombined their efforts to promote generic drugs, which had the same effect but wereconsiderably cheaper. This was not an easy task, because generic drugs were virtuallyunknown in the CFA zone and because powerful lobbies were opposed to the import ofgeneric drugs which were expected to reduce marketing margins. Thanks to this strategy,the cost of essential drugs hardly increased in CFA francs from 1993 to 1998 in most

20 The promotion of generics in some CFA countries started in the early 1990s, but the devaluation provideda new impetus. The promotion campaign was officially launched in a meeting of the health ministries of theCFA countries held in April 1994 in Brussels.

19

CFA countries. This provides an example of successful partnership among donors (Box2.3).

Box 2.3 Generic Drugs and Central Stores: C8te d'Ivoire, Burkina Faso and Mali

On the eve of the devaluation, many public health services were lacking essential drugs, becausethe government agencies responsible for importing the drugs were virtually bankrupt. The first step wasto restructure the agencies administratively, financially (by clearing debts) and physically (byrehabilitating stores). The second step was to provide the seed capital needed to make the systemfinancially viable by enforcing a cash-and-carry policy and financing two rounds of drug purchases. Thefirst round paid for the initial stock of drugs and the second for the working capital needed to replenishthe stock. By selling drugs with a small mark-up, health centers were expected to cover their currentexpenditures. The system was to become self-financing and it worked reasonably well along the lines ofthe "Bamako Initiative."

In Cote d'Ivoire, which received strong support from the EC, sales by the government central storereached about US$2 Imillion in 1998, of which 34 percent for generics, 22 percent for specialties, 18percent for vaccinations and injections, 14 percent for small equipment, and 12 percent for other items.Metropolitan Abidjan accounted for about 40 percent of the sales. In spite of the devaluation, the cost ofthe package of essential drugs increased by only 13 percent in nominal terms from 1993 to 1998, whichmeans that costs fell by some 25 percent in relation to the purchasing power of the population. InBurkina Faso, sales by the central store reached US$7 million in 1998 and the use of generics progressedrapidly; according to customs data, imports of generics rose from 8 percent of all drug imports in valueterms in 1994 to 25 percent in 1997 and sales of generics by the private sector now exceed those by thepublic sector. The central stores have not been allowed to supply the "for profit" private sector, becauseit was feared that this sector would attempt to disrupt the scheme by acquiring the entire stock ofgenerics. But the existing scheme is not without leakage and part of the drugs imported by the centralstore end up with the "for profit" private sector. The prohibition to sell to the "for profit" sector mayhave been justified in the initial stage, but this prohibition should be progressively removed and thecentral stores should ultimately be privatized.

In Mali, a parastatal (the "Pharmacie Populaire du Mali" PPM) retained a monopoly on theimports of medicine and pharmaceuticals through the early 1990s. After the system was liberalized in1994 in the context of a Bank project, PPM started purchasing generics under international competitivebidding and the cost of the package of essential drugs declined in spite of the devaluation. According to a1998 Bank report, generics accounted for a larger share of drug sales in Mali than in most other WestAfrican countries in 1997.

2.20 It was initially thought that public health service centers were poorly attendedbefore the devaluation because patients were unable to get the drugs they needed. It wasexpected that, once drugs became available and centers rehabilitated, attendance wouldimprove. But this did not happen. As a matter of fact, in several countries, attendancedeclined at public centers, while it increased at private ones. In Burkina Faso, attendanceat public centers declined steadily since 1986. The decline was generally attributed to thepoor quality of services delivered by government employees. Moreover, with theintroduction of cost-recovery, the quality/price ratio moved against public centers.

2.21 In Mali, where the private practice of medicine was illegal until 1985, thepercentage of the national budget devoted to health steadily declined from 9 percent in1970 to 4 percent in 1987 and stagnated in the 5 percent range through 1994. l Sincepublic health services hardly existed in rural areas, health community centers came to fillup the vacuum. A federation of community health associations was created in 1994 and,

21 "The World Bank and the Health Sector in Mali", OED Report 18112, 6/30/1998.

20

with the support of the Bank and other donors, 300 community health centers had beenestablished by 1998. Attendance was greater at community centers than at thosemanaged by the public administration. But community centers offered neither jobsecurity nor pension, which made it difficult to retain staff. Moreover, since salarieswere determined by the communities, the staff was induced to work in the moreprosperous areas where the pay was better.

2.22 In Cote d'Ivoire, the bright spot is the recent establishment of community healthcenters in the Abidjan metropolitan area. The head and treasurer of the centers wereelected by the community and they were responsible for hiring the staff. Becauseattendance was much higher than in government centers, the authorities expressedinterest in community centers. But a top-down approach may not be appropriate toreplicate a successful experiment which started from the bottom. Moreover, the legalstatus of community centers remains ambiguous.

2.23 Cameroon is one of the CFA countries where the quality of public health servicesdeteriorated most dramatically. Doctors employed in government centers consider thatthey are underpaid and they devote most of their energy in building up a private practiceon the side. Although doctors are better paid in private centers, the cost per patient islower because attendance is much higher.

2.24 The percentage of children immunized increased from 1992-93 to 1997-98 in 9 ofthe 13 CFA countries and in 8 of the 12 non-CFA counties for which data are available(Table C5.2). Among CFA countries declines were recorded in Congo, Gabon and Togo.Although improvement was, on average, greater in CFA than non-CFA countries, rates ofimmunization in 1997-98 remain lower in CFA countries. Immunization rates in 1992-93were lower in CAEMC than in WAEMU and the differential widened in 1997-98. Butchanges in immunization rates have to be interpreted with great caution in view ofdiscrepancies between different statistical sources.

2.25 Assistance by the Bank and Other Donors. While the first health project in CFAcountries was financed only in 1983, during the last five years, Bank commitments havebeen higher for health than for education (8.1 percent of total commitments versus 5.7percent, Table C 15). This rapidly increasing involvement of the Bank was not withoutproblems. As noted by OED (footnote 13), "The overarching recommendation of thereview is that the Bank should seek to do better, not more." "Paradoxically, Bank projectdesigns were usually more complex -with a greater number of components andorganizational units-in countries with weak institutional capacity." Moreover, progresswas often assessed in terns of inputs (budget share allocated to health or nurnber of newpublic health centers), instead of results (increase in the number of visits to public healthcenters and in the percentage of the population vaccinated). In January 1999, the Bankhad 13 active health projects in CFA countries and half of the amounts committed in theseprojects was at risk, compared with only one quarter for non-health projects (TableC 16.3). By contrast, for the group of 28 non-CFA countries, the relative amounts at riskwere about the same for health (31 percent) and non-health projects (30 percent).

22 Memorandum to the Executive Directors and the President and Executive summary paragraph 4 in"Development Effectiveness in Health, Nutrition and Population" 5/10/1999, OED Report 19226.

21

2.26 The European Commission (EC) started financing health projects on a significantscale in 1987 and became more involved in 1990 with its new structural developmentfacility. Health accounted for 36 percent of EC' s total budgetary assistance to the CFAzone in 1991-97 and assistance for health increased by 42 percent (in current ECU) from1991-93 to 1994-97. But the EC encountered implementation problems. In particular,during the course of a routine control in Cote d'Ivoire, the EC identified major leakageand decided, at the beginning of 1999, to interrupt disbursement of all budgetaryassistance to the country, pending the implementation of appropriate corrective measures.

2.27 Donors did not adjust fast enough to the rapidly changing structure of the healthsector. Through the early 1980s, basic health services were expected to be deliveredessentially free of charge by the State and non-profit organizations. With the crisis, itbecame clear that the State did not have the resources to fulfill these expectations and theprinciple of cost-recovery became widely accepted.23 Because public health centers werenot anymore free while the quality of services was generally lower than in privatecenters, attendance to public centers declined in many areas. The share of healthexpenditures financed by the State also declined; in Cameroon, it fell to 11 percent. InMali, which is one of the poorest countries, 17 percent of health expenditures werecovered by the State in 1997, 31 percent by donors and the remaining 52 percent byNGOs and households. In Cote d' Ivoire as in Mali, only 20 percent of doctors join thecivil service after graduation, which means that the non-government sector is already thedominant one or soon will be. The State has to remain responsible for the provision ofessential services which the private sector will not provide (such as reproductive health,vaccination campaigns, HIV prevention and services to the very poor). But theseservices do not need to be delivered by civil servants; they may be contracted to privateagencies, if those are able to deliver services more efficiently. The State has to facilitateservice deliveries by the private sector, while regulating private activities in the publicinterest. This requires basic changes in the training and the philosophy of health officialswhich will only occur progressively.

2.28 Since IDA resources are scarce and donors other than the Bank are increasinglyinterested in the health sector, the Bank has to concentrate on what it can do best. TheBank has no comparative advantage over NGOs and specialized agencies (such as WHOor UNICEF) in traditional health projects aiming at the delivery of specific healthservices. But it may have a comparative advantage in assisting governments in findinghow the relationships between the public and the private health sectors should evolve.The attention of donors, Bank included, has been too narrowly focused on public sectordeliveries and too little is known on the health sector as a whole.

2.29 With the new HIPC initiatives, savings from debt forgiveness have to be used toimprove basic health and education. For this purpose, IMF and Bank programs oftenspecify the increase in the budget shares which have to be devoted to these sectors; butsuch simple criteria can be misleading for two reasons. First, an increase in budgetaryallocation does not mean that the money will reach the intended beneficiaries. Second,

23 With the "Bamako Initiative", basic health services were to be delivered by public centers which wereexpected to finance most of their current expenditures through cost-recovery. The donor community was toprovide the seed capital needed for current operations and the resources required to construct (or renovate)public health centers.

22

even if the resources reach the intended beneficiaries, public expenditures may not beused efficiently; what matters is achieving better results, not spending more money. Tothis end, the bank should assist governments in improving the collection and monitoringof basic social statistics. Better statistics are needed to assess the developmenteffectiveness of public expenditures. The Bank should focus its attention on a smallnumber of indicators which can be monitored.

23

3. World Bank Assistance

Financial Assistance from the World Bank and from other Sources2 4

3.1 The following developments characterized the external financial assistance to theCFA zone since 1985:

3.2 (i) From FY85-88 to FY93, gross disbursements to the 13 CFA countries fell by60 percent in relation to gross disbursements to the 28 non-CFA countries.

3.3 (ii) The World Bank accounted for a modest share of official developmentassistance (ODA) to CFA countries. During the period 1985-97, net disbursements bythe World Bank (IDA and IBRD combined) to the 13 CFA countries accounted for 15percent of net ODA disbursements from sources other than IDA, and net transfers foronly 7 percent. France accounted for 41 percent, EC for 12 percent, Germany for 8percent and Japan for 5 percent. The share of France increased throughout the crisis,rising from 28 percent in 1985 to 48 percent in 1993.

3.4 (iii) Even if official development assistance to the CFA zone increased during thecrisis, the increase was much lower than the terms of trade loss. From 1985-86 to 1991-93, the 13 CFA countries gained 1.6 percent of GDP from external assistance,25 but lost8.6 percent of GDP due to the deterioration in the terms of trade. During the sameperiod, the 28 non-CFA countries gained 5.7 percent from external assistance and lost 3.4percent on account of lower terms of trade, which resulted in a net gain equivalent to 2.3percent of GDP. This partly explains why non-CFA countries did better than CFAcountries during this period.26

3.5 (iv) Net external transfers from all sources (official or private, concessional ornot) to the CFA zone expressed as percentage of GNP doubled on an annual basis from1990-93 to 1994, but declined sharply afterward (Figure 3.3). For the zone as a whole,net external transfers fell from 5.2 percent in 1990-93 to 3.3 percent of GNP in 1995-98;for Cameroon and CMte d'Ivoire net external transfers became negative in 1995-98. Thisis the context within which the Bank lending strategy has to be evaluated.

The Bank Lending Strategy

3.6 After recognizing the need for a devaluation in 1988, the Bank reduced itslending to CFA countries, but did not stop lending. One of the main purposes of Banklending in 1988-93 was to provide debtor countries with the breathing space needed to

24 This section summarizes the analysis presented in the first section of Annex A. Country AssistanceEvaluations are available for 4 of the 13 CFA countries: Togo (OED, 11/97), Cote d' Ivoire (Report 19422,OED, 6/14/99), Burkina Faso (OED, 2000) and Cameroon (OED, June 2000).25 Defined as net ODA disbursements from sources other than IDA plus net transfers from IBRD and IDAcombined.26 From 1985-86 to 1991-93, GDP declined by 2.8 percent a year in CFA countries against only 0.9 percentin non-CFA countries, Table 2.1.

24

strengthen their repayment capabilities. But, since in 1987-93 the interest rate on the newdebt exceeded the growth rate of export earnings the problem only worsened. For ParisClub creditors, this led to repeated re-schedulings27 and a series of debt reductioninitiatives with terms becoming progressively more generous (Toronto 1988, Houston1990, Naples 1993, HIPC 1997 and EHIPC 1999). For the World Bank, it led to therepayment of the IBRD debt by IDA credits.

3.7 Stopping new lending would have led to large net transfers to the World Bankfrom Cameroon and, especially, from C6te d'Ivoire (about 2.5 percent of GDP) preciselywhen they were suffering enormous terms of trade losses. This did not appear right for adevelopment agency and it could have raised the risk of default. But committing US$1.5billion to Cote d'Ivoire and Cameroon under IBRD terms in FY86-90 (with half of it asadjustment lending) was not right either, since their creditworthiness at the time wasdoubtful and the projects thus financed did not turn out well.

3.8 In FY86-90, IBRD committed US$1 billion to C6te d'Ivoire for 13 projects andUS$0.5 billion to Cameroon for 8 projects (Table C10). The performance of the 13 IvoryCoast projects was poor: only 41 percent of the amounts committed was rated withsatisfactory outcome and 8 percent with likely sustainability and substantial institutionaldevelopment impact. The performance of the 8 Cameroon projects was even worse: noneof the projects was rated with satisfactory outcome and, of the amounts committed, 90percent was rated with unlikely sustainability and 88 percent with negligible institutionaldevelopment impact. The situation was not the same when the Bank committed US$450million to Cote d'Ivoire under three adjustment loans in FY92, because the loans weredesigned to support the expected devaluation. Moreover, only US$200 million (half ofwhich as IDA credits) were disbursed before the devaluation. It had been specified thatthe remaining US$250 million could not be disbursed before restoring competitiveness,which was understood to mean before devaluating the currency and this understandingwas respected. Still, net World Bank transfers to C6te d'Ivoire remained negative for tenconsecutive fiscal years (FY89-98) and those to Carneroon for seven consecutive fiscalyears (FY92-98). At that time, France argued that external assistance to countriessuffering major terms of trade losses should be raised and not reduced. The Bank on theother hand, argued that a devaluation was needed and that greater financial assistancewould only delay the unavoidable. At times relations with France became quite tense.Exerting pressure while avoiding open confrontation was the difficult role played by theBank for several years. Beginning in 1988, IMF and Bank staff discussed the measuresneeded to prepare for a devaluation with senior officials in the CFA zone. In July 1992,the IMF Managing Director undertook a mission to selected CFA countries to discuss adevaluation; but several Heads of State were not ready to devalue. In March 1993, theways in which the Bank could support a CFA devaluation were reviewed with Frenchtreasury officials. On January 10, 1994, the Heads of State of the CFA countries wereinvited to meet in Dakar under the pretext of reviewing the problems of Air-Afrique.Two days later, the devaluation was announced to the public.

27 Over a ten-year period (1981-91) Senegal went nine times to the Paris Club and Togo went eight times.

25

3.9 Immediately after the devaluation, Cameroon, Congo and C6te d'Ivoire weredeclared eligible only for IDA. IDA disbursed US$585 million after the devaluationunder the three projects which had been negotiated before the devaluation. IDA was thusable to disburse US$150 million to C6te d'Ivoire in the month following the devaluationand the World Bank has to be credited for this timely IDA intervention which contributedto the success of the devaluation. The IBRD investment portfolio was not, however,restructured rapidly enough. Most investments loans were allowed to complete theircycle under IBRD terms and the last IBRD investment loan was closed in FY99. Incountries with poor portfolio performance such as Cameroon, it would have beenpreferable to cancel more loans.

3.10 Commitments from the World Bank to the CFA zone increased sharply after thedevaluation, but net transfers did not because of the heavy servicing of the IBRD debt.As a matter of fact, average annual net transfers in current US dollar were 13 percentlower in FY94-99 than in FY90-93 (Figure 3.1). When net transfers are expressed aspercentages of GNP instead of current US dollar, the picture appears somewhat morefavorable, because the GNP of CFA countries fell in US dollar terms after thedevaluation. World Bank net transfers to the CFA zone increased from 0.2 percent ofGNP 1990-93 to 0.4 percent in 1995-98, while IMF net transfers increased from -0.4percent to +0.4 percent (Table A3.1). However, due to large repayments of arrears toAfDB which had been accumulated before the devaluation, net transfers from allmultilaterals (including the IMF) remained unchanged at 0.3 percent of GNP (Figure3.2). It is clear that policies of CFA countries were better in 1995-98 than in 1990-93,but the countries were not rewarded for this improvement by multilateral developmentagencies.

26

Fig. 3.1: World Bank Commitment and Net Transfer to the 13 CFA Countries, FY90-99(in millions of current US $)

r-1,200

Commitment eorganization o Afnca De t ___1,000 __________

2000

-200

90 91 Fiscal 92 93 94 95 96 97 98 99

Fig. 3.2: Net Transfers to the 13 CFA countries from Multilateral Creditorswith and without IMF, 1985-98

(as percent of GNP)

3

2.5 -- - Including

20 5-58_B_8B9 9 _1 92_ 4 5 W 9

A.~~~

from mutlaea Exclud ecldig Ing 989

C ~~~~~~~~A,

*26 1 - --7 -- -V U\

.0.5

0 -- ----

-0.5 -

-1

85 86 87 88 89 90 91 92 93 94 95 96 97 98

Fig. 3.3 Net Transfers to the 13 CPA countries from all soulres andfrom multilateral creditDrs excluding IMF, 1985-98

(as percent of GNP)

80 iŽLTrnsfer, long-term loans tMF ____ AEL purchases & grantsz06

02

I. -X - -- - --

-2 85 86 87 88 89 90 91 92 93 94 95 96 7977 98~

Official and private long-term creditors at concessional and non-concessional terms, IMF and grants

other than technical assistance 2rants. See Annex A Section 1.3. oamra2rrhs 12-14 and Tables A3 and A4

27

Non-Lending Assistance

3.11 The quality of Bank lending and non-lending services is often affected by the sizeof the administrative budget. The share of Africa in the Bank's administrative budget fellfrom 36 percent in FY90 to 25.5 percent in FY99. During the same period, the share ofCFA countries in Africa increased only marginally, because most of the gain in WAEMU(from 14.2 percent to 16.7 percent) was offset by a loss in CAEMC (from 8.2 percent to6.2 percent). The administrative budget of the Africa region valued in real termsremained almost unchanged from FY90 to FY94, but fell by 16 percent from FY94 toFY99 (a fall larger than in any other region) with declines of 5 percent for WAEMU and33 percent for CAEMC (Table C12). The budget constraint became tighter in FY00 andthe situation is expected to become more difficult in FY01. The reduction in theadministrative budget has probably become the major problem in improving the qualityof Bank assistance to CAEMC countries.

3.12 Expenditures28 on Bank economic and sector work managed by the Africa region(CESW) fell even more. From FY94 to FY98, CESW expenditures valued in real termsfell by 56 percent for CAEMC and 49 percent for WAEMU, against 49 percent for Africaas a whole and 37 percent Bank-wide (Table C13). Some economic and sector work mayhave been conducted in the context of lending operations, but such work has a relativelynarrow focus since it has to be closely related to the specific objectives of the loan. Atthe same time, according to the QAG which evaluated 40 CESW issued by five regions inFY98, Africa had the lowest percentage of satisfactory reports (55 percent against 72percent Bank-wide).

Performance Ratings

3.13 Since country performances have been rated by the Bank every year since 1977,Bank ratings provide the obvious starting point to assess the performances of CFAcountries. For each of the four periods retained in this report, CFA countries wereclassified into six groups. Equatorial Guinea was the CFA country with the most steadyrating; it remained in the bottom group throughout the 22 year period. Cameroon andCBte d'Ivoire were the two CFA countries with the most dramatic changes in rating.They remained in the top group through 1986, but fell down to group 3 in 1987-93. Inthe post-devaluation period (1994-97), Cote d'Ivoire moved up to group 5 (the secondfrom the top) while Cameroon moved down to group 2 (the second from the bottom). Inretrospect, CMte d'Ivoire in 1977-80 and Cameroon in 1981-86 should not have beenrated in the top group. If country performance were to reflect the efforts made by thecountry, CMte d'Ivoire should be rated better in 1994-98 than in 1977-80 and Cameroonbetter in 1994-98 than in 1981-86.

28 Expenditures financed from budget and trust fund resources combined. Expenditures financed from trustfunds increased sharply in FY94, but fell in FY97-98 to less than one-quarter of the FY94 peak.Expenditures shown in Table C12 do not include Trust Fund. Moreover, they are adjusted backward to thereduction of benefits (from 70 percent to 50 percent) introduced in FY99, while expenditures growth in TableC13 are not since they do not include FY99.

28

Projects

3.14 The composition of Bank lending changed considerably during the last 20 yearsand changes were greater in CFA than in non-CFA countries. The share of investmentprojects in agriculture and infrastructure fell from 83 percent of total Bank commitmentsbefore FY81 to 26 percent after FY94 in CFA countries, compared with a decline from73 percent to 44 percent in non-CFA countries (Table C 15). The sharp decline in CFAcountries was compensated by increases from 0 to 8 percent for health projects and from0 to 53 percent for adjustment lending.29

3.15 The quality of projects, as rated by OED, was affected by the economicenvironment, especially by the fiscal situation which prevailed during implementation.For projects approved through FY83, the proportion of projects with satisfactory outcomewas higher for CFA than non-CFA countries (72 percent versus 59 percent, Table C 16.2).But the reverse was true for projects approved during the FY84-89 period (36 percentversus 61 percent), with the greatest deterioration in CAEMC (Table C17 and FiguresC 1.1 and C 1.2). In the FY99 portfolio, the percentage of projects at risk (weighted bycommitments) was lower in WAEMU (16 percent) than in the 28 non-CFA countries (30percent) and, especially, in CAEMC (77 percent, Table C16.3). In the case of healthprojects, however, risk was higher in WAEMU (48 percent) than in the non-CFA group(31 percent).

World Bank Performance

3.16 During the period ending in 1980, Bank performance was broadly satisfactory.OED project ratings3 0 were significantly higher for CFA than non-CFA countries foroutcome, and marginally higher for sustainability and institutional development (TableC1 7). The Bank did not succeed in convincing governments to cool down overheatedeconomies, but the Bank's main objective at the time was to promote good investmentprojects.

3.17 During the second period (1981-86), Bank performance was marginallysatisfactory in WAEMU and marginally unsatisfactory in CAEMC. OED project ratingsremained higher for WAEMU than for non-CFA countries for both investment andadjustment projects; but public enterprise restructuring was not successful, although itwas supported by large Bank lending. CAEMC countries did not receive any adjustmentlending and the three oil countries, which were booming, were not receptive to Bankpolicy advice. In the climate of euphoria, poor investment decisions were made and,even for projects financed by the Bank, OED ratings of investment projects felldramatically, far below average ratings for non-CFA countries.

3.18 During the third and the fourth periods, the rating is more complex. Taking ashort-term view, Bank performance would be rated unsatisfactory in 1987-93 andsatisfactory in 1994-98, which is consistent with OED rating on adjustment lending. In

29 Of total adjustment lending to the zone, 43 percent went to CMte d'Ivoire, 15 percent to Cameroon, 12?ercent to Senegal and the remaining 30 percent was shared among the ten other CFA countries.

Average weighted by commitment size.

29

particular, in the case of adjustment loans approved by the Board in the FY86-89 period,outcome was rated satisfactory in only 25 percent of the cases for CFA countries (45percent for WAEMU and 2 percent for CAEMC), against 61 percent for non-CFAcountries; ratings for sustainability and institutional development were even worse. Bycontrast, for adjustment loans approved in FY94-95, outcome was rated satisfactory in 86percent of the cases for CFA countries. Recent improvements are also noticeable.Quality at entry was marginally lower for CFA than for non-CFA countries in FY97, butit became marginally higher in FY98. Regarding rapid supervision assessment, WAEMUwas rated equal to non-CFA countries in FY97 and marginally better in FY98, butCAEMC remained below the non-CFA average.

3.19 Taking a longer-term view, it is not possible to draw a clear demarcation linebetween Bank performances before and after devaluation day. The Bank contributed tothe success of the devaluation, not only by what it did after devaluation day, but also bywhat it did before.

30

4. Support to Regional Integration

4.1 Although the Bank nonnally conducted its work at the country level, it followed aregional approach to restructure the WAEMU banking sector in the late 1980s. This ledto a study of the competitive position of the entire CFA zone and the preparation ofcontingency programs to support the expected devaluation. The regional approach wasrelevant and efficacious in these two cases. It has become relevant again with the presentdrive toward regional integration, notably with the recent decision made in six ECOWAScountries to establish a new monetary union which could later be merged with WAEMU.

Table 4.1: ECOWAS GNP, GDP, Population 1997-98

GNP' GDP2 Population GNP GDP Population

US$ billions Millions Percentage of totalBenin 2.3 2.2 5.9 3.1 2.8 2.7Burkina Faso 2.6 2.5 10.6 3.5 3.1 4.8C6te d'lvoire 10.2 10.7 14.4 13.8 13.4 6.5Mali 2.7 2.6 10.5 3.6 3.3 4.8Guinea-Bissau 0.3 0.3 1.2 0.3 0.3 0.5Niger 2.0 2.0 10.0 2.7 2.5 4.5Senegal 4.7 4.6 8.9 6.4 5.7 4.1Togo 1.5 1.5 4.4 2.0 1.9 2.0WAEMU 26.2 26.2 65.7 35.5 33.0 29.9Nigeria 34.8 40.7 119.3 47.2 51.2 54.3Ghana 7.2 7.2 18.3 9.8 9.1 8.3Guinea 3.9 3.8 7.0 5.3 4.7 3.2Sierra Leone 0.8 0.7 4.8 1.0 0.9 2.2Liberia .. .. 3.0 .. .. 1.3Gambia, The 0.4 0.4 1.2 0.5 0.5 0.5Cape Verde 0.5 0.5 0.4 0.7 0.6 0.2Other ECOWAS 47.5 53.2 153.9 64.5 67.0 70.1

Total 73.7 79.4 219.6 100.0 100.0 100.0Sub-Saharan Africa 326.1 339.4 619.4TAtlas Method;2 at market prices;

4.2 The Bank, together with the IMF, assisted BEAC (the central bank of CAEMCcountries) and COBAC (the banking commission) in the restructuring of the financialsector. Starting five years before the devaluation, the Bank advised UDEAC (the CentralAfrican Customs and Monetary Union, which has been replaced by CAEMC) indesigning a major tariff and tax reform which was implemented in 1994. But, while theregional tariff reform was in place earlier in CAEMC than in WAEMU, the CAEMCtreaty was ratified in June 1999, almost five years after ratification of the WAEMUtreaty. Due to this delay, external assistance in the post-devaluation period was focussedon the WAEMU Commission.

31

4.3 France, the EU, the Fund and the Bank have been the main partners of theWAEMU Commission. The EU has been supporting the operating budget of thecommission; it also financed numerous studies, workshops and ad hoc technicalassistance. France provided technical assistance and contributed to the financing ofspecialized regional institutions, such as AFRISTAT and OHADA. The Bank and theFund, in close cooperation with the EU and France, have been advising the WAEMUcommission in the establishment of a common external tariff (which is in place sinceJanuary 2000) and on other aspects of trade policies. They have also assisted thecommission in harmonizing indirect taxes, formulating a "competition policy" anddesigning a common investment code (which has not yet been ratified). The Bank,jointly with the Fund, became involved in the area of multilateral surveillance (publicfinance in particular). IDA and IFC assisted WAEMU in establishing a regional stockexchange to improve capital mobility within the sub-region. The regional exchangeopened in Abidjan in September 1998.

4.4 When its mandate was renewed in January 1999, the WAEMU commission wasasked to focus its work on sectoral policies; particularly, as regards agriculture, energy,transportation and telecommunications. The Bank responded by giving attention totelecommunications and power interconnections not only within WAEMU, but alsobetween WAEMU and Nigeria and Ghana. The Bank is well equipped to assist theWAEMU commission in reviewing sectoral policies at the regional level and suchreviews could benefit both WAEMU and the Bank. Reviews should also cover the socialsectors (poverty reduction, education and health); such reviews would provide theopportunity of establishing and monitoring sets of social indicators comparable amongcountries of the sub-region. Regional workshops could be organized on specific issueswith representatives of the countries and the Commission. Their purpose would be toidentify missing or unreliable information, to determine what works and what does not,and to understand why. Once appropriate and reliable social indicators become available,the Commission could select some of them as convergence criteria to monitor socialprogress in the Union.

4.5 The Bank could assist the regional commissions efficiently in areas where it hascomparative advantages, notably sectoral reviews and relations with the outside. Withinthe Bank, a regional approach could generate synergies and cross-fertilization amongcountry teams. It could also provide a framework for harmonizing country assistancestrategies, thus improving the effectiveness of country-based work.

32

5. Conclusions and Recommendations

5.1 The CFA arrangement induced a fiscal discipline which was beneficial throughthe mid 1980s; afterward, it introduced a rigidity which was detrimental until it wasrelaxed by the 1994 devaluation. Because foreign exchange and monetary policies aredetermined at the supra-national level, the burden on national fiscal policies isparticularly heavy. It became unbearable in the second half of the 1 980s with thedeterioration of the terms of trade and the appreciation of the CFA franc. In hindsight, itis clear that delaying the devaluation was very costly in economic and social terms. Butthe Bank was not responsible for the postponement, and it is not clear that a successfuldevaluation could have taken place earlier if the Bank would have acted differently.When it occurred in January 1994, it was well prepared and it was successful. The Bankreacted promptly, and during the first year it was able to support the best performers withsignificant increases in financial assistance on IDA terms.

5.2 Poverty assessments had the merit of drawing public attention to the povertyproblem, but their efficacy was limited. Assessments have to become more actionoriented and they have to be better integrated in the overall country strategy; this is thepurpose of the PRSP now under preparation in most CFA countries. Strategies to supportrural development are needed since the majority of the poor live in rural areas. Thesecould become the conerstone of the PRSPs.

5.3 Setting up minimum budget shares to priority sectors can be misleading unless thenature and the impact of public expenditures are closely monitored. To monitor progressand assess development effectiveness, the Bank needs reliable indicators expressed interms of results. The Bank should assist in the collection of reliable statistics and thedevelopment of indicators to monitor improvements in these sectors and the effectivenessof its assistance.

5.4 The Bank is right in intensifying its efforts to contain the spread of HIV. In viewof the complexity and the magnitude of the task, the Bank has to work very closely withthe other agencies which can bring a contribution. An example of successful partnershipwas the introduction of generic drugs through central public stores. These stores weregenerally not allowed to sell drugs to the private profit-oriented sector. This prohibitionmay have been justified in the initial stage; but it is a source of leakage and it should beprogressively eliminated.

5.5 The attention of donors, Bank included, has been too narrowly focused on thedelivery of health services by the public sector. More attention should be given todelivery by non-government agencies, which is the sub-sector expanding most rapidly.The Bank appears better positioned than other agencies to assess the appropriate balancebetween service delivery by government and non-government agencies and to analyze thelinks between the health sector and the rest of the economy.

5.6 Bank assistance followed a regional approach only in a few cases, and thisapproach has been fruitful. With the present drive toward regional integration, thisapproach would be fruitful again.

33

5.7 The Bank has a comparative advantage in assisting the WAEMU commission intwo areas: first, assessing the relations of WAEMU with the outside world and, inparticular, with the six ECOWAS countries aiming at establishing a new monetary union;second, within WAEMU, reviewing national policies by sectors. Moreover, within theBank, a regional approach could generate fruitful synergies among country-teams; itcould also provide a useful framework for harmonizing country assistance strategieswithin the region.

5.8 The Bank should play a more active role in reviewing sectoral policies at theregional level; notably, as regards agriculture, energy, transportation andtelecommunications, which are of particular interest to the WAEMU commission. TheBank should also assist the commission in: (i) organizing workshops on key social issuesin order to determine what works or does not work in the region and to understand why;and (ii) selecting realistic and monitorable social targets which could be later adopted bythe Commission as social convergence criteria.